Making a BD139 voltage follower buffer, need a little help

[FallenAngel]

Hey guys,

I was curious into the Heed CanAmp in the way it uses a single NPN transistor (BD139-16) as a voltage follower as the buffer to an opamp, so I decided to make a much nicer amp but include this buffer.

I have a dual regulated power supply (very similar to 2 of Tangent's TREADs) which to power this thing.

The amp circuit is the same opamp amplifier circuit used by the Pimeta/PPAv2 (implements Jung's multi-loop topology, without CCS on the opamp).

The buffer circuit is a BD139-16 voltage follower biased into Class-A using a 100 Ohm / 5 Watt resistor with a 220 Ohm resistor on output (in the feedback loop of the opamp through a 10K resistor).

The schematic looks like this:

I'm getting an output DC of 600mV which drops to 500mV over 10 seconds or so, but hangs around there afterwards.

Any idea on why this is happening, why the Heed CanAmp uses this buffer without DC offset and what can do to fix it?

Thanks!

 

[TzeYang]

erm are you sure you can multi loop this? The HEED IIRC uses conventional single set of feedback resistors to control the gain and set feedback. Plus, there is a capacitor to ground at the feedback loop (this method makes sure that DC isnt amplified.)

 

[FallenAngel]

Well, I tried with multiloop and single loop but the offset is exactly the same. I did notice the Heed having a capacitor to ground following the 1K resistor, and noticed the same in the GainClone I'm building, but didn't try that.

That's probably what it's there for. I'll see what I can do with it. I don't think I have any high quality caps to go there, so I might have to go with a Panasonic FM. Any idea on what size it should be? The Heed (and a similar BD139 buffer) shows 10uF, the GainClone shows 22uF and general GainClone discussion suggest 47uF. I'll have to check what I have.

I think I have a pair of 470uF BlackGate Standard series, 4.7uF Nichicon Muse ES and a pair of 47uF BlackGate NX but they are only 6.3V.

 

[majkel]

You built too complictaed and DC-uncompensated feedback loop, that's why you get offset voltage.
1) Remove parts: 1M, 475k, 3.32k
2) Replace 4.32k for 910 ohm
3) Replace 200 ohm resistor for wire.
4) When strange behavior occurs, add capacitor parallel to 10k resistor, starting with 22pF. If still unstable - increase the capacity, if stable - try to lower the capacity as much as possible.

 

[FallenAngel]

Majkel:

The 1M to ground on input is to set the impedance and I think it might be a good idea to keep it so the source doesn't see 0-50K which changes with volume setting.

475K and 3.32K are for multiloop - I can definitely remove those and make it single loop.

I'll change the 4.42K on the input to 1K (I don't have a 910 Ohm handy).

I'll see how this works, hopefully there won't be much "strange behavior".

What do you think about adding the capacitor in the feedback loop?

 

[majkel]

You can place 10k parallel to 1k in place of the 910 ohm, and then will be perfect. 1k is probably still OK, just observe the offset voltage. Your explanation for 1M resistor is invalid. The source always gets 50k of load, whatever setting of volume is applied. If it makes you feel better, leave the 1M resistor in it's place. BTW, what op-amp are you using? I guess at such a high gain =11 there won't be any instability and the additional capacitor won't be necessary.
One more thing: use a pot of lower impedance, for instance 10k, then it will be easier to compensate the offset voltage when you use op-amps with high input current (read: many bipolar).

 

[TzeYang]

ah yes, bipolars. But since this is quite an experiment, would it not be safer to use fet input opamps first? (fallenangel did not specify which opamp but i was just making an assumption lol)

 

[NelsonVandal]

Tori amp with BJT
HeadWize: DIY Workshop > Changing the Tori amp to use a BJT

 

[TzeYang]

the circuit actually works. I did a simulation just to be sure. offset is extremely low. Of course, without jung multiloop and OPA627 as opamp.

 

[amb]

Yup, you can't multiloop this design because the opamp's output needs to be Vbe=~0.6V higher than the BD139 emitter (where you want it to be 0). The inner loop would force the DC offset up.

 

[~n00beR]

Not of a multi loop design however, after examining the pictures of the amp (google images) I came to the attached schematic.

It seems to simulate fine in TINA however, im new to using the software.

Any comments suggestions welcome

 

[FallenAngel]

New schematic following (most of) Majkel's advice

Opamp in there was an OPA2227 (which might have fried during initial testing since offset was 600mV), but after replacing it with OPA2132 and OPA2134 the offset is 150mV and 200mV respectively. Still definitely not a good thing, but better. Just to test, I tried thowing in a pair of OPA637 - the offset is 300mV with those.

I don't think the 200 Ohm resistor is the culprit (at least I hope not).

Any ideas on what else I could try (except for adding resistor parallel to 10K (I don't have a very small one handy, just 0.1uF X7R Ceramic)

Also, if going with the capacitor to ground after the 1K resistor (I heard this is a bad thing for audio in general), what voltage rating does it need to be?

 

[~n00beR]

In the canamp, the capacitor to ground in the feedback loop is an electrolytic, 10uF/100V

I think the main problem is that you are missing the input capacitor.

I have simulated the effects of with and without the input capacitor, and it needs to be there.

The use of the capacitor to ground in the feedback does effect the offset, however as far as i can tell the offset will still be low enough for headphones without it.. However if both are not there, as we both now know damage will be done.

Input cap is 0.47uf non polar.

Hope this is of some help

 

[amb]

Why do you have that 200R resistor at the output anyway? The reason for the BD139 being there is to serve as a current booster, but all that benefit is lost on that resistor... If you must use a resistor there, it should be very low value, not hundreds of ohms.

I also suggest adding a series resistor at the base of the BD139, so that the opamp isn't driving the base directly. Something like 47-100 ohms should do. It will help prevent parasitic oscillations. I also suggest a small value capacitor going from the opamp output to its inverting input (see the M³ amp schematic, or the Pimeta's ground channel as example). This should also help with stability.

Inserting a cap from the bottom of the 1K resistor to ground will give 100% NFB at DC and will surely reduce output DC offset, but you will need a large value electrolytic... 10uF will yield a high-pass -3dB frequency of around 16Hz in this circuit. This cap should normally "see" very little voltage but you should use something that is rated at least the rail voltage just in case. Also, it should be a non-polar cap.

 

[~n00beR]

ahhh, miss read the begining of the thread
Yes, listen to amb